Fine-scale soil mapping in Alaska, traditionally relying on fieldwork and localized simulations, remains a critical yet underdeveloped task, despite the region's ecological importance and extensive permafrost coverage. As permafrost thaw accelerates due to climate change, it threatens infrastructure stability and key ecosystem services, such as soil carbon storage. High-resolution soil maps are essential for characterizing permafrost distribution, identifying vulnerable areas, and informing adaptation strategies. We present MISO, a vision-based machine learning (ML) model to produce statewide fine-scale soil maps for near-surface permafrost and soil taxonomy. The model integrates a geospatial foundation model for visual feature extraction, implicit neural representations for continuous spatial prediction, and contrastive learning for multimodal alignment and geo-location awareness. We compare MISO with Random Forest (RF), a traditional ML model that has been widely used in soil mapping applications. Spatial cross-validation and regional analysis across Permafrost Zones and Major Land Resource Areas (MLRAs) show that MISO generalizes better to remote, unseen locations and achieves higher recall than RF, which is critical for monitoring permafrost thaw and related environmental processes. These findings demonstrate the potential of advanced ML approaches for fine-scale soil mapping and provide practical guidance for future soil sampling and infrastructure planning in permafrost-affected landscapes. The project will be released at https://github.com/knowledge-computing/Peatland-permafrost.

Retrieval Augmented Generation (RAG) has become prevalent in question-answering (QA) tasks due to its ability of utilizing search engine to enhance the quality of long-form question-answering (LFQA). Despite the emergence of various open source methods and web-enhanced commercial systems such as Bing Chat, two critical problems remain unsolved, i.e., the lack of factuality and clear logic in the generated long-form answers. In this paper, we remedy these issues via a systematic study on answer generation in web-enhanced LFQA. Specifically, we first propose a novel outline-enhanced generator to achieve clear logic in the generation of multifaceted answers and construct two datasets accordingly. Then we propose a factuality optimization method based on a carefully designed doubly fine-grained RLHF framework, which contains automatic evaluation and reward modeling in different levels of granularity. Our generic framework comprises conventional fine-grained RLHF methods as special cases. Extensive experiments verify the superiority of our proposed \textit{Factuality-optimized RAG (FoRAG)} method on both English and Chinese benchmarks. In particular, when applying our method to Llama2-7B-chat, the derived model FoRAG-L-7B outperforms WebGPT-175B in terms of three commonly used metrics (i.e., coherence, helpfulness, and factuality), while the number of parameters is much smaller (only 1/24 of that of WebGPT-175B). Our datasets and models are made publicly available for better reproducibility: https://huggingface.co/forag.

There is a growing academic interest as well as commercial exploitation of millimetre-wave scanning radar for autonomous vehicle localisation and scene understanding. Although several datasets to support this research area have been released, they are primarily focused on urban or semi-urban environments. Nevertheless, rugged offroad deployments are important application areas which also present unique challenges and opportunities for this sensor technology. Therefore, the Oxford Offroad Radar Dataset (OORD) presents data collected in the rugged Scottish highlands in extreme weather. The radar data we offer to the community are accompanied by GPS/INS reference - to further stimulate research in radar place recognition. In total we release over 90GiB of radar scans as well as GPS and IMU readings by driving a diverse set of four routes over 11 forays, totalling approximately 154km of rugged driving. This is an area increasingly explored in literature, and we therefore present and release examples of recent open-sourced radar place recognition systems and their performance on our dataset. This includes a learned neural network, the weights of which we also release. The data and tools are made freely available to the community at https://oxford-robotics-institute.github.io/oord-dataset.

Knowledge probing assesses to which degree a language model (LM) has successfully learned relational knowledge during pre-training. Probing is an inexpensive way to compare LMs of different sizes and training configurations. However, previous approaches rely on the objective function used in pre-training LMs and are thus applicable only to masked or causal LMs. As a result, comparing different types of LMs becomes impossible. To address this, we propose an approach that uses an LM's inherent ability to estimate the log-likelihood of any given textual statement. We carefully design an evaluation dataset of 7,731 instances (40,916 in a larger variant) from which we produce alternative statements for each relational fact, one of which is correct. We then evaluate whether an LM correctly assigns the highest log-likelihood to the correct statement. Our experimental evaluation of 22 common LMs shows that our proposed framework, BEAR, can effectively probe for knowledge across different LM types. We release the BEAR datasets and an open-source framework that implements the probing approach to the research community to facilitate the evaluation and development of LMs.

Can you relate to this problem? I can't find anything to play on iOS right now. Now don't worry mobile developers--it's not you, it's me. I don't go for puzzle games or narrative titles because I want something more mindless before I sleep, or while I'm on the bus. I just have a different problem: I've played you already.

Artificial intelligence--AI--is an information technology. And it is already deeply embedded into our social fabric, both in ways we understand and in ways we don't. It will hack our society to a degree and effect unlike anything that's come before. I mean this in two very different ways. One, AI systems will be used to hack us. And two, AI systems will themselves become hackers: finding vulnerabilities in all sorts of social, economic, and political systems, and then exploiting them at an unprecedented speed, scale, and scope. We risk a future of AI systems hacking other AI systems, with humans being little more than collateral damage. Okay, maybe it's a bit of hyperbole, but none of this requires far-future science-fiction technology. I'm not postulating any "singularity," where the AI-learning feedback loop becomes so fast that it outstrips human understanding. My scenarios don't require evil intent on the part of anyone. We don't need malicious AI systems like Skynet (Terminator) or the Agents (Matrix). Some of the hacks I will discuss don't even require major research breakthroughs. They'll improve as AI techniques get more sophisticated, but we can see hints of them in operation today. This hacking will come naturally, as AIs become more advanced at learning, understanding, and problem-solving. In this essay, I will talk about the implications of AI hackers. First, I will generalize "hacking" to include economic, social, and political systems--and also our brains. Next, I will describe how AI systems will be used to hack us. Then, I will explain how AIs will hack the economic, social, and political systems that comprise society. Finally, I will discuss the implications of a world of AI hackers, and point towards possible defenses. It's not all as bleak as it might sound. Caper movies are filled with hacks. Hacks are clever, but not the same as innovations. Systems tend to be optimized for specific outcomes. Hacking is the pursuit of another outcome, often at the expense of the original optimization Systems tend be rigid. Systems limit what we can do and invariably, some of us want to do something else. But enough of us are. Hacking is normally thought of something you can do to computers. But hacks can be perpetrated on any system of rules--including the tax code. But you can still think of it as "code" in the computer sense of the term. It's a series of algorithms that takes an input--financial information for the year--and produces an output: the amount of tax owed. It's deterministic, or at least it's supposed to be.

Evaluating and quantifying the beauty of a landscape, an ecosystem and its effects on a person's well-being has become a central issue for public authorities. With this in mind, scientists from the École Polytechnique Fédérale de Lausanne (EPFL) in Switzerland and Wageningen University in the Netherlands have developed a new indicator based on deep learning and several million photos posted on the social network Flickr. An article was recently published in Nature Scientific Reports. When we walk in nature, whether in the mountains, in a forest or by the sea, we feel things, a certain well-being. Numerous studies have highlighted the benefits of such activities for our health, both physical and mental.

Using regression kriging Malone et al. (2011) implemented the PI The Brazilian territory covers an 8.5 million km The Amazon biome covers almost half of Brazil (49.29%), with a

On the occasion of my sixtieth birthday, my friend Lenny visited me from Toronto. He is seven years older than me, and he gave me some sound advice: respect the limitations of your body. Lenny said that he no longer climbs ladders, even though he is a yoga instructor and his balance is good--climbing ladders just seems like a risky thing for a sixtysomething to do. The advice came just after I had binge-watched the first season of "Westworld," a TV series about machines gaining human consciousness (something that I, like many cognitive neuroscience professors, have been teaching for over ten years). In the world of the show, the bodies of the robots, unlike your body and mine, are easily repaired. A vast robot-repair shop remanufactures and reattaches severed limbs, and efficiently closes gaping wounds. For the past few years, I've been on a kick that I call the "pre-mortem": thinking ahead to what could go wrong and putting systems in place to minimize the damage if they do go wrong. For instance, I got a landline, in case the cell networks go down in a natural disaster such as an earthquake. I've taken cell-phone photos of my passport and credit cards, in case they get lost. I taped an emergency-phone-number list to the inside of the kitchen cabinet that is nearest the phone, and I put a combination-lock box in the back of my house to hold a front-door key, in case I lock myself out. I must have struck a chord with this idea, because my TED talk about it went viral. My wife, Heather, and I have our bedroom upstairs, and there is only one way out in case of a fire--down the stairs and out the front door.

This research hypothesized that a practical approach in the form of a solution framework known as Natural Language Understanding and Reasoning for Intelligence (NaLURI), which combines full-discourse natural language understanding, powerful representation formalism capable of exploiting ontological information and reasoning approach with advanced features, will solve the following problems without compromising practicality factors: 1) restriction on the nature of question and response, and 2) limitation to scale across domains and to real-life natural language text.